Major labs currently withhold internal models stronger than public releases.
@BlackHC honestly I don't really understand the question. The labs have stronger internal models, yes I know this. I think it's good that there's competition that'll drive a need for broader access.
@natolambert What about all the strong models you never had access to?
How would you trade off not having access at all vs the issues you're discussing here?
Sorry this was in regards to "Don't release the model if you can't hit your safety targets."
I don't think more competition will drive broader access necessarily for these use-cases
I would expect other big AI labs to do the same (re degredation) but communicate it better
(Speaking in personal capacity and having no insights etc)
@BlackHC honestly I don't really understand the question. The labs have stronger internal models, yes I know this. I think it's good that there's competition that'll drive a need for broader access.
Major labs currently withhold internal models stronger than public releases.
@BlackHC honestly I don't really understand the question. The labs have stronger internal models, yes I know this. I think it's good that there's competition that'll drive a need for broader access.
@natolambert What about all the strong models you never had access to?
How would you trade off not having access at all vs the issues you're discussing here?